Editor’s Note: Today we’re featuring an article from Joe Marchese, co-founder of Human Ventures and a longtime advertising executive, about what advertising in generative AI might look like, and why it’s important to get it right. Joe’s also joined our private discord server, and you can participate the discussion about this story via the link at the bottom of the article. – Alex Kantrowitz
You’re halfway home from work when you decide to have a spontaneous early dinner at the new fast-food place in your neighborhood. You were going to hit the gym, but you thought “why not treat myself?” so you tell your AI assistant to reschedule gym time for tomorrow.
What you don’t know is that you are one of the 100,000 people the company that owns the assistant has promised would try Acme Burger that week. You are the product, and the money you spend on the burger is the ROI of a performance marketing campaign.
If you think this is far-fetched, stop reading now. You haven’t been paying enough attention to the shift from “advertising” to “performance marketing.” Advertising promises to deliver a message in return for human attention. Performance marketing promises outcomes. And promising outcomes has an entirely different meaning when it comes to generative AI and the large language models behind it.
Advertising will shape generative AI platforms as it has shaped every medium of communication prior. And that could have a hugely disruptive effect on search, and the trillions of dollars of business built on top of it. Not to mention the entire content ecosystem which has fed search for the past 20+ years and is feeding generative AI today.
I’m an advocate for the advertising industry, but only to a very specific point. Brands subsidizing our stories and experiences in exchange for the ability to deliver their message is a trade-off people instinctively understand and can freely accept. We allow brands to borrow our most precious resource, our time and attention, in return for promises of a quality product or experience. But brands also need to deliver on those promises or the bargain falls apart.
The entire online economy is built on top of systems trying to capture and optimize our time and attention. The generative AI explosion kicked off with a seminal paper at Google, and used by ChatGPT, that’s based on something called the “attention mechanism” which reflects the way that humans process language. The paper this idea was founded on was called “Attention Is All You Need.”
The most effective advertising gives people transparency and agency over how their attention is being monetized. But when we abdicate the optimization for our attention to algorithms and AI, it can lead to massive problems.
What Google did brilliantly with search, better than any before, was ensure that advertising was additive to the search experience. There was, and still is, utility in search ad results. And paid search results are well marked as advertisements, in order to maintain trust, lest people start using alternative search engines.
The most effective advertising gives people transparency and agency over how their attention is being monetized. But when we abdicate the optimization for our attention to algorithms and AI, it can lead to massive problems.
Poor advertising experiences can kill entire mediums. Just look at cable TV. It’s not the content, or the bundle that bothers most people (I’ll happily debate this with anyone), it is the viewer’s experience. And cable isn’t the only medium where the consumer experience was entirely broken by the advertising business model.
With the trillions of dollars at stake in the generative AI wars, platforms will not, or perhaps I should say should not, risk losing people’s trust or degrading the experience for advertising revenue. But if history is any indication, the platform that wins in generative AI will figure out how to make advertising additive to user the experience, transparent, and, perhaps most importantly, support the quality content ecosystem that generative AI platforms need to maintain relevance, namely publishers of quality content.
This means that publishers of original content should be more valuable than ever. The publishers in particular should benefit from this dynamic because what is generative AI without taste, and how relevant is generative AI if it’s not up to speed on the news?
A truly sustainable business model for LLMs should value platforms like Reddit, LinkedIn, YouTube or X, and publishers like Conde Nast, Dotdash Meredith, The New York Times, Reuters and Forbes, which deliver verified facts and foster human conversation. It’s ironic that the editorial authority that defines traditional publisher brands is exactly what generative AI needs for fresh, relevant content, yet many of those same outlets are the most under pressure from the collapse of traffic.
Creating value through fair advertising has always been the trick and could be the key to a generative AI advertising system that benefits platform and publisher partners. But it suggests a positive path forward for generating AI advertising: one that’s transparent, values human attention, and is actually useful, and isn’t just making promises to marketers that we’ll eat more burgers without realizing why we’re doing it.
Here’s some good news: ChatGPT actually agrees with me on this, or at least it says it does. I asked about the risk to consumers of hidden advertising in LLMs. Its response? “If a generative AI system like ChatGPT is also an ad platform, its primary job becomes not helping you, but monetizing you.”
This article is from Big Technology, a newsletter by Alex Kantrowitz.
The post Ads Are Coming to AI and None of Us Are Ready appeared first on TheWrap.